mapspinner 0.1.50 → 0.1.52

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "mapspinner",
-  "version": "0.1.50",
+  "version": "0.1.52",
   "description": "WebGL2 Earth-scale terrain rendering SDK for interactive globe applications",
   "main": "src/index.js",
   "exports": {

package/src/gl-render.js

@@ -788,7 +788,12 @@ export async function initMapspinnerRender(gl, opts = {}) {
     // extends the horizon to tens of km at low altitude so near-shore relief stays in view (negligible
     // depth-precision cost: 500m vs R~6.37e6). Both the cull and the draw use this (single source).
     const RHORIZON = R - 500.0;
-    const horizon = Math.sqrt(Math.max(0.0, camDist*camDist - RHORIZON*RHORIZON));
+    // UNDERWATER FAR-PLANE FIX (user 2026-06-14 'at -214m visible, at -500m it disappears'): when the
+    // camera is more than 500m below sea level, camDist < RHORIZON so the sea-level horizon is imaginary
+    // (-> 0) and alt is negative; the old max(horizon, alt*8) then collapsed the far plane to ~0 and the
+    // whole scene vanished past -500m deep (= the 'ocean looks shallow/empty' when exploring). Floor the
+    // far reach to 60km when submerged so the seabed + the underwater view stay visible.
+    const horizon = (camDist > RHORIZON) ? Math.sqrt(camDist*camDist - RHORIZON*RHORIZON) : 60000.0;
     // MATCH render()'s near exactly (2026-06-14 jank fix): the cull frustum must use the SAME near
     // as the draw frustum, else behind-limb/screen-AABB culling diverges from what is actually drawn
     // at the deck (cull near was max(*0.1,0.1) while render used the <2m 0.05 branch).
@@ -831,7 +836,12 @@ export async function initMapspinnerRender(gl, opts = {}) {
     // extends the horizon to tens of km at low altitude so near-shore relief stays in view (negligible
     // depth-precision cost: 500m vs R~6.37e6). Both the cull and the draw use this (single source).
     const RHORIZON = R - 500.0;
-    const horizon = Math.sqrt(Math.max(0.0, camDist*camDist - RHORIZON*RHORIZON));
+    // UNDERWATER FAR-PLANE FIX (user 2026-06-14 'at -214m visible, at -500m it disappears'): when the
+    // camera is more than 500m below sea level, camDist < RHORIZON so the sea-level horizon is imaginary
+    // (-> 0) and alt is negative; the old max(horizon, alt*8) then collapsed the far plane to ~0 and the
+    // whole scene vanished past -500m deep (= the 'ocean looks shallow/empty' when exploring). Floor the
+    // far reach to 60km when submerged so the seabed + the underwater view stay visible.
+    const horizon = (camDist > RHORIZON) ? Math.sqrt(camDist*camDist - RHORIZON*RHORIZON) : 60000.0;
     const near = altAboveTerrain < 2.0 ? 0.05 : Math.max(altAboveTerrain * 0.1, 0.05);
     const _fBlend = Math.min(1.0, Math.max(0.0, (alt - 500000.0) / 4500000.0));
     const farGround = Math.max(horizon, alt * 8.0);

package/src/shaders/terrain.glsl

@@ -1391,7 +1391,13 @@ vec3 terrainAlbedoClimate(float h, float slope, float rockSlope, float temp, flo
         return mix(c, vec3(0.82, 0.88, 0.94), seaIce * 0.9);
     }
     // biome weight: full on gentle lowland, fading where rock/snow/steep takes over.
-    float veg = (1.0 - smoothstep(bandEdgesHi.x, bandEdgesHi.y, h)) * (1.0 - smoothstep(slopeRock.x, slopeRock.y, slope));
+    // VEG SLOPE GATE (user 2026-06-14 'hard line of light grass around rocks'): the gate faded veg over
+    // `slope` (the macro, gentler) while the ROCK blend keys off `rockSlope` (the raw geometric normal,
+    // steeper). The two thresholds didn't coincide, so on the approach to a rock the dark biome-green
+    // vegetation dropped out a beat BEFORE the rock arrived -> a light bare-grass ring. Key veg off the
+    // SAME rockSlope so vegetation fades EXACTLY as rock fades in -> grass stays dark right up to the
+    // rock with no light gap, and veg is 0 wherever rock is full (no biome-green bleeding onto rock).
+    float veg = (1.0 - smoothstep(bandEdgesHi.x, bandEdgesHi.y, h)) * (1.0 - smoothstep(slopeRock.x, slopeRock.y, rockSlope));
     // ELEVATION + LATITUDE BIOME BIAS (user 2026-06-02: 'the hypsometric ramp should influence biome
     // distribution, pulling biomes out of their anchor areas a bit for better distribution'). On top
     // of the anchor climate temp/humid, bias the EFFECTIVE temperature DOWN with elevation (lapse rate